Atypical Listeria innocua strains possess an intact LIPI-3

Background Listeria monocytogenes is a food-borne pathogen which is the causative agent of listeriosis and can be divided into three evolutionary lineages I, II and III. While all strains possess the well established virulence factors associated with the Listeria pathogenicity island I (LIPI-1), lineage I strains also possess an additional pathogenicity island designated LIPI-3 which encodes listeriolysin S (LLS), a post-translationally modified cytolytic peptide. Up until now, this pathogenicity island has been identified exclusively in a subset of lineage I isolates of the pathogen Listeria monocytogenes. Results In total 64 L. innocua strains were screened for the presence of LIPI-3. Here we report the identification of an intact LIPI-3 in 11 isolates of L. innocua and the remnants of the cluster in several others. Significantly, we can reveal that placing the L. innocua lls genes under the control of a constitutive promoter results in a haemolytic phenotype, confirming that the cluster is capable of encoding a functional haemolysin. Conclusions Although the presence of the LIPI-3 gene cluster is confined to lineage I isolates of L. monocytogenes, a corresponding gene cluster or its remnants have been identified in many L. innocua strains.


Background
Listeria monocytogenes is a food-borne pathogen which is the causative agent of listeriosis [1][2][3][4][5]. It has long been known that the characteristic haemolytic phenotype of L. monocytogenes is attributable to the activity of listeriolysin O (LLO), encoded by the hly gene located within Listeria Pathogenicity Island I (LIPI-1) [5]. However, more recently, it has also been revealed that several strains of lineage I L. monocytogenes (of four evolutionary lineages, serotype 4b strains within lineage I have been most commonly associated with outbreaks [6]) (also possess an additional pathogenicity island (designated LIPI-3) which encodes a second haemolysin, designated listeriolysin S [7][8][9]. Listeriolysin S (LLS) is not normally expressed in vitro, and hly mutants give a non-haemolytic phenotype on blood agar. LLS is one of a growing number of posttranslationally modified cytolysins (post-translationally modified haemolytic peptides) that include the Streptococcus pyogenes-associated Streptolysin S (SLS) and the Clostridium botulinum/Clostridium sporogenes-associated Clostridiolysin S and is a member of the broader family of thiazole/oxazole modified microcins (TOMMs) [9]. It has been established that LLS plays a role in the survival of L. monocytogenes in PMNs and also contributes to virulence in the murine model [8]. LIPI-3 consists of 8 genes arranged in the following order: llsAGHXBYDP. LlsA is the structural peptide; LlsB, Y and D are enzymes proposed to perform the post-translational modifications; LlsGH is an ABC transporter; LlsP is a protease; while LlsX is of unknown function [7,8]. The associated promoter, P llsA , which is situated upstream of llsA, is not transcribed in standard laboratory media but is induced by oxidative stress. It has been suggested that expression of the LIPI-3 genes may be induced in the phagosome of macrophages [8]. When P llsA is replaced by a constitutive promoter (P HELP ), a strongly haemolytic/cytolytic phenotype is revealed under laboratory conditions [8]. The inducible nature of LLS and its absence in many L. monocyctogenes strains is probably responsible for the fact that this virulence factor has gone undetected until recently.
Listeria innocua is an avirulent species within the Genus Listeria. It has been proposed that L. innocua and L. monocytogenes have evolved from a common ancestor and differ predominantly due to the loss of virulence genes by L. innocua [10,11]. This is supported by the existence of atypical L. innocua isolates which retain LIPI-1 and other virulence factors [12,13]. In a previous investigation we demonstrated that none of 11 L. innocua isolates examined (one of which was initially classified as an L. grayi isolate) possessed the equivalent of the LIPI-3 [7,8]. In this study we extended our analysis to a larger collection of strains, which has revealed that several strains possess the remnants of a LIPI-3. In fact, 11 strains possess fully intact LIPI-3 which gives rise to a haemolytic phenotype when the genes are constitutively expressed.

Strains and growth conditions
Tables 1, 2, and 3 list the panel of Listeria strains used in this study. Strains were obtained from the Food Microbiology Microbial Collection (University College Cork) and the Special Listeria Culture Collection (SLCC). All strains were cultured at 37°C for 16 h in Brain Heart Infusion (BHI) broth or agar (Oxoid, Hampshire, UK) unless otherwise stated. Where necessary, the characterisation of strains as L. innocua was confirmed biochemically by means of the API listeria kit (BioMérieux, Lyon, France) and 16S ribosomal DNA (rDNA) with CO1 and CO2 primer pairs previously described by Simpson et al. [14]. Escherichia coli EC101 was used as an intermediate vector host. Antibiotics were incorporated as follows [8]: Erythromycin (Ery) 150 μg/ml E. coli, 5 μg/ml L. innocua. Chloroamphenicol (Cm) 10 μg/ml E. coli and L. innocua. Ampicillin (Amp) 100 μg/ml E. coli. 5-bromo-4-chloro-3indolyl-b-D-galactopyranoside (X-Gal) was incorporated at a concentration of 40 μg/ml.

Sequence analysis
A PCR-based strategy, employing the primer pair llsAFor-llsARev, was employed to screen for the presence of the LLS structural gene, llsA. These and other primers corresponding to regions both within (1113for, 1114rev, 1115 rev, 1118rev, 1120rev) and surrounding (araCrev) the LIPI-3 of L. monocytogenes F2365 were employed to amplify flanking DNA sequences which were subsequently sequenced (MWG Biotech) ( Table 4). Primer Lin1080_F1, which was designed to amplify from the conserved gene, corresponding to lin1080 in strain CLIP11262, was used to determine the position of LIPI-3 in L. innocua strains relative to this locus. Overlapping sequences were assembled and a consensus sequence was determined using  Constitutive expression of the LIPI-3 cluster of L. innocua strain FH2051 The L. innocua FH2051 lls genes were placed under the control of the strong constitutive synthetic promoter P HELP using the pORI-based repA-negative plasmid system as previously described by Cotter et al., with some modification [8]. Briefly, P HELP DNA was amplified with the primer pair PhelpFsoe(LI)/PhelpRsoe from the plasmid pPL2luxPHelp [16] and fused between two DNA fragments amplified from the regions flanking P llsA by splicing by overlap extension (SOE) PCR [17]. The upstream region was amplified with the primer pair PllsAchgA(LI) and PllsAchgB(LI) and the downstream region was amplified with primers PllsAchgC and PllsAchgD. All PCRs were performed using Vent DNA polymerase (NEB, New England Biolabs, MA, USA). The SOE PCR product was cloned into the multiple cloning site (MCS) of pORI280 following PstI and EcoRI (NEB) digestion and ligation with the Ligafast rapid DNA ligation system (Promega, Madison, USA). The sequence of the cloned product was verified with MCS primers pORI280For/Rev by MWG Biotech, Germany [18]. Pellet-paint (Novagen) precipitated plasmid was subsequently transformed into the intermediate repA-positive host E. coli EC101. The plasmid was co-transformed into L. innocua FH2051 with the highly temperature-sensitive plasmid pVE6007 which supplies RepA in trans. Transformed cells appeared as blue colonies following plating on BHI-Ery-Xgal at 30°C. The integration of pORI280 by single crossover homologous recombination was stimulated by picking a single blue colony from the transformation plate and incubating it on BHI-Ery-Xgal at 30°C for 24 h and subcultured twice on BHI-Ery-Xgal at 42°C. A second crossover event, resulting in the introduction of P HELP in place of P llsA and the eventual loss of the pORI280 vector, was screened for following multiple subcultures in the absence of antibiotic selection. The introduction of P HELP upstream of llsA in Ery resistant Cm sensitive colonies was confirmed by PCR. A haemolytic phenotype was determined by spotting 10 μL of an overnight culture of this strain onto Columbia blood agar (Oxoid) containing 5% defibrinated horse blood (TCS Biosciences, Buckingham, UK) and 1 mU/ml sphingomyelinase (Sigma) and examining after 24 h.

Pulsed-field gel electrophoresis
Pulsed-field gel electrophoresis was carried out following the CDC standardized PulseNet protocol for  L. monocytogenes with AscI and ApaI as the restriction endonucleases. The PFGE patterns were analyzed using BioNumerics software [19].

Results and discussion
Screening L. monocytogenes and L. innocua for homologues of llsA To date LIPI-3 has been identified in~60% (27 of 46) of lineage I L. monocytogenes but was absent from all lineage II (n = 23) and lineage III (n = 5) isolates tested [8]. As a consequence of gaining access to the Seeliger collection of Listeria isolates [20], we were provided with the opportunity to screen for the presence of LIPI-3 among an additional 83 L. monocytogenes isolates including 30 lineage I, 50 lineage II and 3 lineage III strains. The llsA gene was not identified in any lineage II or lineage III strain, consistent with our previous observations (Table 1). However, the llsA gene was identified in 70% of lineage I L. monocytogenes screened (21 of 30) and, on the basis of PCR amplification, in all cases the full complement of LIPI-3 genes was present. All such isolates originated from human, animal (including milk and feed) and sewage sources. When collated with data from previous studies, it is apparent that 63% (48 of 76) of lineage I isolates are LIPI-3 positive and may be capable of LLS production. All LIPI-3 positive isolates belonged to Lineage I as verified by an allele specific oligonucleotide PCR multiplex (actA1-f, actA1-r, plcB2-f, plcB2-r, actA3-f, plcB3-r) based on the prfA virulence gene cluster [15], thus verifying previous observations with respect to the distribution of LIPI-3 among different evolutionary lineages of L. monocytogenes [7,8].
Access to the Seeliger collection and other strains also facilitated a further investigation of the LIPI-3 status of L. innocua. As stated, a previous analysis of 11 strains of L. innocua indicated that all lacked genes associated with LIPI-3 [7,8]. However, screening a larger collection of 64 L. innocua strains using llsA specific primers revealed that 45 strains (70.3%) were llsA-positive (Table 3). Further PCR-based analysis of these isolates, employing a variety of primers designed to amplify across and within the LIPI-3 (llsAFor, llsARev, 1113for, 1114rev, 1115rev, 1118rev, 1120 rev, araCrev) revealed that 11 of these strains possess a cluster which is comparable in size, gene content and gene organisation to that of the LIPI-3 cluster found in a subset of lineage I L. monocytogenes strains. These 11 isolates originated from a number of European countries between 1984 and 2000, and were isolated from varied sources including processed chicken [1], cheese [7], sheep [7], silage [7] and human [1] (Table 3). Further analysis revealed that 25 L. innocua isolates possess a truncated LIPI-3 with no PCR product generated for llsBYDP. Sequencing the region confirmed that these genes are absent in at least two isolates (SLCC6270 and SLCC6382). With the exception of llsP, these genes have previously been found to be essential for LLS production in L. monocytogenes [7]. Of the remaining 28 strains, 9 were found to contain llsA but attempts to amplify across or within other LIPI-3 associated genes were unsuccessful and another 19 isolates lacked all LIPI-3 genes. Two L. innocua isolates, SLCC6382 and SLCC6270, containing a truncated LIPI-3, were selected for further analysis. Both SLCC6382 and SLCC6270 shared 98% homology with respect to the structural peptide LlsA. The putative LlsG, LlsH and LlsX proteins from both strains shared 96%, 99% and 95% identity with their L. monocytogenes counterpart. llsB, llsY, llsD and llsP are absent from both isolates, while the AraC-like regulatory protein determinant was present with 98% identity to the L. monocytogenes cluster. As in L. monocytogenes, the L. innocua cluster is located downstream of a putative glutamine hydrolyzing GMP synthase protein (GuaA). However, the island in SLCC6382 and SLCC6270 commences 600 bases immediately downstream of guaA and thus is not flanked by glyoxylase encoding genes, thereby contrasting with LIPI-3 in L. monocytogenes.
Three strains (SLCC6466, SLCC6294, FH2051) possessing an entire LIPI-3 cluster were also selected for a more extensive investigation. Eight complete ORFs were identified, each corresponding to their homologue in the L. monocytogenes LIPI-3 cluster (llsAGHXBYDP). Sequence alignments confirmed considerable homology at the protein level (Figure 1). The structural peptide LlsA shared  98% homology in the case of the three strains mentioned above to the L. monocytogenes equivalent. These L. innocua clusters also encode homologs of the putative two component ABC transport system LlsG and LlsH, with LlsG sharing 95.3% (FH2051) and 95% (SLCC6466, SLCC6294) identity, and 98.8% (FH2051) and 99% (SLCC6466, SLC C6294) with respect to LlsH. The putative LlsX homolog, which is of unknown function, is 97% identical to its L. monocytogenes counterpart for all three isolates. This gene is believed to be specific to LIPI-3 since no homologue exists among other sag-like gene clusters [7]. A corresponding cluster of putative Lls homologs, all of which are predicted to encode biosynthetic enzymes, were also identified [8]; LlsB (99% in the case of all three strains), LlsY (95.4% FH2051, 95% SLCC6466 and SLCC6294) and LlsD (98.4% FH2051, 98% SLCC6466 and SLCC6294). Finally, the L. innocua cluster also carries putative LlsP and Lmof2365_1120 homologs, annotated as a CAAX aminoterminal putative metalloprotease and AraC-like regulatory protein which share 93.8% FH2051, 91% SLCC6466 and SLCC6294 and 91.3% FH2051, 94% SLCC6466 and SLCC6294 identity to the L. monocytogenes cluster, respectively. PFGE was carried out to assess the relatedness of the 11 L. innocua strains harbouring intact LIPI-3 a s. On the basis of this analysis, all LIPI-3 + isolates share a high degree of similarity, with the majority of strains (SLCC6466, SLCC6814, SLCC6749, SLCC6276, SLCC6279, SLCC6294, FH2051, SLCC6296 and SLCC6298) displaying 80% similarity and strains SLCC6203 and SLCC7199 sharing 76% identity ( Figure 2). The LIPI-3 + L. innocua FH2051 is non-haemolytic when grown on Columbia blood agar (Figure 1). This is not surprising given that L. innocua strains do not produce LLO and the fact that it has previously been established that LLS is not produced by wild type L. monocytogenes in laboratory media. It has been established that the latter is due to the fact that P llsA is not transcribed under standard laboratory conditions [8]. It has been noted previously that P llsA is induced under oxidative stress but, unfortunately, the requirement for an oxidizing agent prevents an assessment of associated haemolytic activity on blood agar [7]. Thus, to investigate the functionality of the LIPI-3 cluster in L. innocua, here we constitutively expressed LIPI-3 through the introduction of the constitutive Highly Expressed Listeria Promoter [P HELP , (LLS C )] upstream of llsA in L. innocua FH2051, to create FH2051LLS C . Examination of the resultant strain revealed that the L. innocua LIPI-3 is indeed functional as evidenced by a clear haemolytic phenotype on Columbia blood agar ( Figure 3).

Conclusion
In conclusion, we have established that although the presence of the LIPI-3 gene cluster is confined to lineage I isolates of L. monocytogenes, a corresponding gene cluster or its remnants can be identified in many L. innocua. It is now generally accepted that L. innocua and L. monocytogenes evolved from a common ancestor, with L. innocua having lost virulence genes since this division. Although rare, L. innocua isolates exist which possess the LIPI-1 gene cluster and another L. monocytogenes associated virulence gene, inlA [12,13]. Nonetheless, the retention of the LIPI-3 cluster by a large proportion of strains is unexpected. The LIPI-3 clusters in the various L. innocua strains seem to be at various stages of reductive evolution with a number of stains possessing an intact island, others showing clear evidence of disintegration and yet another group in which the island is completely absent. It is not clear, however, whether the gradual loss of LIPI-3 from L. innocua strains is a slow process that has been underway since the existence of the last common ancestor of Figure 2 Dendrograms derived from PFGE profiles of AscI and ApafI macrorestriction displaying restriction pattern similarity among the 11 L. innocua LIPI-3 + isolates. Figure 3 Growth, after 24 h at 37°C, of L. innocua FH2051 and FH2051LLS C (10 μL spots of an overnight cultures) on Columbia blood agar containing 5% defibrinated horse blood and 1 mU/ml sphingomyelinase.